Real-Time Mapping of Tissue Properties for Magnetic Resonance Fingerprinting

Magnetic resonance fingerprinting (MRF) is a relatively new multi-parametric quantitative imaging method that involves two-step process: (i) reconstructing series of time frames from highly-undersampled non-Cartesian spiral k-space data and (ii) pattern matching using the to infer tissue properties (e.g., \(T_1\) \(T_2\) relaxation times). In this paper, we introduce novel end-to-end deep learning framework seamlessly map directly MRF data, thereby avoiding time-consuming processing such as non-uniform fast Fourier transform (NUFFT) dictionary-based fingerprint matching. Our consumes performs adaptive density compensation, predicts multiple property maps in one forward pass. Experiments on both 2D 3D demonstrate quantification accuracy comparable state-of-the-art methods can be accomplished within 0.5 s, which 1,100 7,700 times faster than original framework. The proposed thus promising for facilitating adoption clinical settings.